Mobility-Aware Traffic Offloading via Cooperative Coded Edge Caching

• Main Authors: Dewang Ren, Xiaolin Gui, Kaiyuan Zhang, Jie Wu
• Format: Article
• Language: English
• Published: IEEE 2020-01-01
• Series: IEEE Access
• Subjects: Cooperative edge caching; network traffic offloading; user mobility; submodular function
• Online Access: https://ieeexplore.ieee.org/document/9023480/

With caching popular contents at the small-cell base stations (SBSs), cooperative edge caching has emerged as an effective approach to offload explosively increasing network traffic from a massive number of users in mobile edge networks (MENs). However, most previous works ignored impact of user mobility on caching strategy, thus having limited practical applications. How to improve the caching strategy by exploiting user mobility is still a challenging problem. Accordingly, in this paper, we propose a mobility-aware and cooperative coded caching (MoCoCoCa) for achieving traffic offloading in MENs, by considering user mobility and hard service deadline constraints. Specifically, we first develop the MoCoCoCa framework to satisfy users' requests locally and conduct the construction of access trajectory set via random walks on a Markov chain, which takes the randomness of contact into account. Then, based on user mobility predictions, we formulate the caching problem into a mixed integer nonlinear programming (MINLP) problem to minimize the load of core network, subject to finiteness of network resources (e.g., storage capacity, transmission rate) and non-uniformity of content popularity constraints. The objective function's submodular property is proved and a cooperative placement algorithm is given, which can achieve (1-1/e) -optimality. Furthermore, to adapt the given content placement configuration to the spatio-temporal dynamics of realtime request flow, the original placement problem is decomposed into some independent subproblems of content placement at individual SBSs. Each subproblem is converted into a submodular optimization problem, and a request-aware distributed replacement algorithm with linear computational complexity is proposed. Trace-based simulations and numerical results demonstrate that our proposed caching strategy can offload up to 69% network traffic than existing caching strategies.